Repeated evaluation of arguments And possible improvements #1

New Issue

2017-05-10T10:14:09+01:00

yjqww6 commented

2017-05-10 10:14:09 +01:00

(Migrated from github.com)

`> (map (begin (displayln 1) cons) '(1 2 3) '(4 5 6))
1
1
1

: (Listof (Pairof Positive-Byte Positive-Byte))
'((1 . 4) (2 . 5) (3 . 6))
`

And, it seems like the 'generalize' function could be replaced by this one?
(: gen (All (a) (case→ [Null → Null] [(Listof a) → (Listof a)]))) (define (gen a) a)

Moreover, is it better to restricted the use of 'set!' to only the end of iteration?

`> (map (begin (displayln 1) cons) '(1 2 3) '(4 5 6)) 1 1 1 - : (Listof (Pairof Positive-Byte Positive-Byte)) '((1 . 4) (2 . 5) (3 . 6)) ` And, it seems like the 'generalize' function could be replaced by this one? `(: gen (All (a) (case→ [Null → Null] [(Listof a) → (Listof a)]))) (define (gen a) a)` Moreover, is it better to restricted the use of 'set!' to only the end of iteration?

SuzanneSoy commented

2017-05-10 14:13:26 +01:00

(Migrated from github.com)

Hi @yjqww6 !

Thanks for the report, I fixed the multiple evaluation problem.

I'm looking at the definition of gen. If I used this double-reverse trick, there probably was a good reason for that, so I'm trying to see if your suggestion brings in any unsuspected regressions.

I'm not sure what you mean by "better to restrict the use of set! to only the end of iteration'? Could you clarify? set! is already used right at the end of each iteration, just before calling the loop function again.

Hi @yjqww6 ! Thanks for the report, I fixed the multiple evaluation problem. I'm looking at the definition of `gen`. If I used this double-reverse trick, there probably was a good reason for that, so I'm trying to see if your suggestion brings in any unsuspected regressions. I'm not sure what you mean by "better to restrict the use of `set!` to only the end of iteration'? Could you clarify? `set!` is already used right at the end of each iteration, just before calling the `loop` function again.

yjqww6 commented

2017-05-10 15:52:34 +01:00

(Migrated from github.com)

@jsmaniac Sorry my for poor explanation. I will paste a example tomorrow.

yjqww6 commented

2017-05-11 03:13:53 +01:00

(Migrated from github.com)

@jsmaniac Okay, I found what I say about 'set!' has a regression on the case of acc with a Rec type. Please forget it.

Sign in to join this conversation.